An Introduction to Bayesian Regression Trees

Abstract

The Bayesian Additive Regression Trees (BART) were first introduced in 2010 and has become an important tool to link traditional Bayesian methods and machine learning techniques. This talk will introduce the method from first principles and show the kind of advantages of using an approach that also considers uncertainty when making predictions. We will also point out some of the disadvantages in current implementations of BART, as well as the possible extensions of the model.

Date
Event
Location
Curitiba, ParanĂ¡, Brazil